Following downregulation of VEGFD in CA1 pyramidal neurons, the cells exhibited shorter and simpler basal dendrites and longer, more complex apical dendrites. The changes in dendritic morphology were particularly prominent in s. oriens, s. radiatum, and s. lacunosum-moleculare, areas where CA1 pyramidal cells receive most of their excitatory input.
These results showed that VEGFD is involved in the in vivo regulation of the dendritic arborization of CA1 pyramidal cells, as already suggested by Mauceri et al. (2011). The experiments of Mauceri and colleagues showed an overall shorter and simpler dendritic arbor in cultured neurons. However, cultured neurons do not have a basal-apical dendritic polarity and are not embedded in the extracellular matrix and in highly organized neuronal networks. Due to more extrinsic factors, the regulatory mechanism of VEGFD in vivo might be more complex than in vitro.
The mechanisms to regulate dendrites are summarized by the term dendritic plasticity. Dendritic plasticity has been observed in the cortex of adult rodents, for example under enriched-environment conditions, sensorimotor learning, chronic exposure to drugs or under pathological conditions. The plasticity of dendrites is important in order to be able to adapt to certain stimuli, but also to maintain the current morphology and the synaptic connections within the network (Hickmott and Ethell 2006).
It has been shown that neurons are able to keep the overall balance of their dendritic trees by homeostatic mechanisms, which can take place on the following dendritic levels: (1) balance between whole dendritic compartments e.g. basal-apical, apical-apical (2) individual trees, originating from the soma and (3) the branches of subtrees (daughters and granddaughters; Samsonovich and Ascoli 2006). Samsonovich and Ascoli concluded that fluctuations in size in a given portion of the neuron (level 1-3) are systematically counter-balanced by the remaining dendrites in the same cell. This suggests further that, a neuron is first interested in keeping the global stability of synaptic input and thus the connectivity in the network. Second, in order to do so, neurons regulate different sub-compartments individually.
Homeostatic mechanisms could be involved in a rearrangement of the dendritic morphology after VEGFD loss. This can result in the shortening of basal dendrites in one compartment, which becomes counter-balanced by longer apical dendrites or vice versa. An unaltered total dendritic length (basal + apical) after VEGFD downregulation supports this theory. The consequence could be the maintenance of the overall synaptic inputs. This hypothesis was confirmed by the stimulation of two input pathways and recording of evoked potentials, which were not altered (Figure 3-16 and section 4.4).
The morphology homeostasis can be influenced and regulated by intrinsic and/or extrinsic factors (Samsonovich and Ascoli 2006). The work by Mauceri et al. and the experiments of this study have identified VEGFD as an intrinsic factor. It is likely that other signal molecules counteract the effects of VEGFD to maintain overall neuronal excitatory status. In particular, the transmembrane protein Lrig1 (leucine-rich repeats and immunoglobulin (Ig)-like domains 1) seems to have complementary effects to VEGFD on dendritic morphology in CA1 pyramidal cells (Alsina et al. 2016). Lrig1 is linked to a tyrosine kinase associated receptor and is able to regulate neurotrophic growth factors like BDNF (Brain-derived neurotrophic factor). Alsina et al. showed an increase in dendritic arborization after knockdown of Lrig1 in cultured neurons. Thus, Lrig1 shows the opposite effect of in vitro knockdown of VEGFD. Interestingly, the in vivo downregulation of Lrig1 in CA1 hippocampal neurons showed only an increase in length and complexity in apical dendrites and no effect on basal dendrites. The authors concluded a preferential regulation of Lrig1 of apical dendrites. Further, Lrig1 knockdown mice have a social interaction deficit. It is conceivable that signal molecules such as Lrig1 interact with VEGFD to regulate and maintain dendritic complexity in different branch types individually to maintain a stable input level to the cell.
The different effects on basal and apical could also result from a non-uniform distribution of factors or mechanisms involved in VEGFD signaling, thus causing distinct effects on basal and apical dendrites. As a secretory protein, VEGFD depends on the Golgi apparatus. It has been shown that the position of the Golgi apparatus determines the basal-apical polarity of neurons (de Anda et al. 2005). Further, the position of Golgi-outposts in dendrites enhances local secretory trafficking. These mechanisms could be involved in the differential regulation of basal and apical dendrites by VEGFD. This was shown for other proteins like the glycoprotein Reelin or the neurotrophic factor BDNF (Matsuki et al. 2010; Horton and Ehlers 2003).
After secretion of VEGFD in the extracellular space, it acts in an autocrine mechanism through VEGFR-3 (D Mauceri et al. 2011). One possible mechanism to achieve a local regulation of dendritic length could be a different VEGFR-3 expression in a certain compartment. The presence of the VEGFR-3 mRNA and expression in somata of CA1 neurons has been shown by in situ hybridization and RT-PCR, respectively (Shin et al. 2008). Furthermore Bhuiyan et al., 2015 showed an increase in VEGFR-3 expression in the soma of CA1 pyramidal neurons and in somata of neurons located in s. radiatum and s. lacunosum-moleculare after ischemia induction. So far, there is no literature about the distribution of VEGFR-3 expression in distinct layers of the CA1 area. Unfortunately, the antibodies for immunostainings of VEGFR-3 are not reliable. The local enrichment of VEGFR-3 seems plausible as other channels and receptors are known to be expressed CA1 layer specific, e.g. HCN 1 and 2 channels, GIRK channels and AMPA receptor (Kupferman et al. 2014).
In cell culture, VEGFD downregulation led to an overall shortening of dendrites (D Mauceri et al. 2011). However, the change in dendritic geometry after the knockdown in vivo was layer specific. This could indicate homeostatic effects based on extracellular factors, e.g. laminated incoming fibers. CA3 neurons project their axons via the Schaffer collaterals both to basal and proximal apical dendrites (Ishizuka, Weber, and Amaral 1990; Spruston 2008). A homeostatic mechanism could take place in order to counter- balance the CA3 input to CA1 neurons. Further, the inhibitory network differentially affects basal and apical dendrites of CA1 pyramidal neurons. The changed dendritic morphology of pyramidal neurons could be a mechanism in order to maintain inhibitory input over the complete dendritic branches. The soma and basal dendrites of OLM interneurons are located in the s. oriens and project their axons to the apical dendrites of pyramidal cells in s. lacunosum-moleculare (Klausberger 2009). OLM cells could provide a mechanism for comparison between compartments and provide information for further morphological homeostasis adjustments.
In order to exclude a bias due to natural variabilities in hippocampus size of animals, the widths of single CA1 layers were measured. The comparison showed no difference between animals and groups. Therefore, a bias can be excluded. To exclude further false positive results by Sholl analysis, the starting point of the Sholl spheres was located at the border between s. radiatum und s. lacunosum-moleculare. This confirmed the finding from the conventional Sholl analysis. However, it can not be fully excluded that the differences in apical dendritic length are due to CA1 cell type differences. In order to avoid this in our dataset, the position of the somata within the pyramidal cell layer was verified and showed no difference between cells. A bias from superficial versus deep pyramidal cells can be probably excluded (Figure 2-4). Due to the precise virus injection by stereotactic coordinates, the recorded cells were in a similar location within CA1, since only cells in the center of injection were approached. Allocated cells or cells with cut dendrites were excluded from the experiment. In summary, bias from hippocampal widths, the location of somata with the pyramidal cell layer and within the whole CA1 area was minimized as much as possible.
In summary. the work by Mauceri et al. and the experiments of this study have identified VEGFD as an intrinsic factor for dendritic morphology regulation in adult mice. Mauceri et al. showed an overall shortening of dendrites in cultured neurons, concluding if VEGFD is suppressed dendrites become shorter. The in vivo data here adds new information to the VEGFD mechanism. The results showed after VEGFD downregulation basal dendrites were shorter and apical longer, but the overall dendritic tree size did not change. Since the cell in vivo still needs to integrate into the network, homeostatic mechanisms could be involved to compensate dendritic size changes.